Query

[owaisali]

Hi i wanna ask that is this possible for Neural Compute stick to run unet model as you hav provided a repo of its optimized version

Reference : https://github.com/IntelAI/unet/tree/master/2D

[Munesh_Intel]

Hi Owais,

Greetings to you. Points 8 and 9 in the following page clearly describe the steps to convert the U-Net model to IR and run the demo using NCS 2.

https://github.com/IntelAI/unet/blob/master/2D/README.md

These steps have been validated using Intel® Distribution of the OpenVINO™ toolkit 2021.2

Regards,

Munesh

[owaisali]

thanks alot done with that error with your help but now facing this one 

2021-01-25 13:11:39.333195: E tensorflow/stream_executor/cuda/cuda_driver.cc:313] failed call to cuInit: UNKNOWN ERROR (303)

if you could help me with this and also i want to change the dataset for this code can any one help me about that 

Regards

[Munesh_Intel]

Hi Owais,

Glad to know that you’ve solved your error.

The new error that you’re seeing is related to NVIDIA driver.

Please check out the following TensorFlow Git issue for probable solutions:

https://github.com/tensorflow/tensorflow/issues/19266

On the other hand, IntelAI/unet repository contains 2D and 3D U-Net TensorFlow scripts for training models using the Medical Decathlon dataset. Intel AI has used these demos as examples of using Intel® Neural Compute Stick 2 for Medical Imaging.

As such, please feel free to explore with other datasets of your choice and leverage Intel hardware in providing a complete, low-cost, portable solution for healthcare AI applications at the edge.

Regards,

Munesh

[owaisali]

Thanks a lot for appreciation and help 

I have successfully executed and run this code.

I am trying to run this code on other datsets .In the whole process the actual confusion and difficulty occured was at the stage of inference .I want to make inference file for my own code and its guide available at your site is quite confusing or i will say complex.Glad to have your help in designing inference file for my own code.

Regards 

Owais Ali 

[Munesh_Intel]

Hi Owais,

The process of integrating the Inference Engine into your application is described in detail in the following link:

https://docs.openvinotoolkit.org/2021.2/openvino_docs_IE_DG_Integrate_with_customer_application_new_API.html

Do check out the following pages for example codes of using the Inference Engine in applications.

https://github.com/openvinotoolkit/openvino/blob/master/inference-engine/samples/classification_sample_async/main.cpp

https://github.com/openvinotoolkit/openvino/blob/master/inference-engine/samples/hello_classification/main.cpp

On the other hand, we would like to know which specific section of our guide is confusing to you?

Regards,

Munesh

[owaisali]

Sir 

i just want to change dataset for this code.i have loaded task 06 lungs dataset

after converting the files to IR format when i run inference i am getting this error 

Data batch channel count (4) does not match filter input channel count (1).

[owaisali]

one more question how can i get the details related to dataset BRATS 

[Munesh_Intel]

Hi Owais,

We notice that you’ve opened a different thread for the 'Data batch channel count (4) does not match filter input channel count (1)' error issue. As such, this issue will be discussed in the following thread:

https://community.intel.com/t5/Intel-Distribution-of-OpenVINO/LOADING-ir-FILES/m-p/1251669

Regards,

Munesh

[owaisali]

The actual problem i am facing here is that the dataset you are using BRATS contains images in .npz format .I need your help in visualizing that dataset. I just want to examine that dataset after which i will be able to fine tune my own dataset over your provided Unet model .I will be highly obliged if you could help me out with this 

Thanks in Advance

[Munesh_Intel]

Hi Owais,

You can use BRATS dataset with braTS and braTS_numpy annotation converters.

https://docs.openvinotoolkit.org/2021.2/omz_tools_accuracy_checker_accuracy_checker_annotation_converters_README.html#supported_converters

If you are using BraTS 2018 dataset, I encourage you to use brain-tumor-segmentation-0001 model.

https://docs.openvinotoolkit.org/2021.2/omz_models_public_brain_tumor_segmentation_0001_brain_tumor_segmentation_0001.html

And if you are using BraTS 2019 dataset, I suggest you use brain-tumor-segmentation-0002 model.

https://docs.openvinotoolkit.org/2021.2/omz_models_public_brain_tumor_segmentation_0002_brain_tumor_segmentation_0002.html

Regards,

Munesh

[Munesh_Intel]

Hi Owaisi,

We are unable to help you on queries related to third party datasets. I would suggest you communicate through the email provided in the following official link for information regarding BRATS 2020 data:

https://www.med.upenn.edu/cbica/brats2020/registration.html

On the other hand, I suggest you refer to the following two GitHub pages for example implementation of U-Net Brain Tumor Segmentation.

https://github.com/Lafite-Yu/BraTS_2018_U-Net

https://github.com/zsdonghao/u-net-brain-tumor

Regards,

Munesh

[Munesh_Intel]

Hi Owais,

This thread will no longer be monitored since we have provided solution and references. If you need any additional information from Intel, please submit a new question.

Regards,

Munesh